On Justifying Epistocratic Political Institutions: A Reply to Samuel Bagg, Cyril Hédoin

Samuel Bagg (2021) has provided a sharp and thoroughly argued reply to my article “The ‘Epistemic Critique’ of Epistocracy and Its Inadequacy” (Hédoin 2021). This article was itself largely motivated by recent contributions intending to reject epistocracy based on what I have called an ‘epistemic critique’. I was especially targeting Julian Reiss’s (2019) article which, in short, argues against the justification of epistocracy on the ground that social scientific knowledge is controversial and cannot help at establishing what superior political judgments are … [please read below the rest of the article].

Image credit: Chupacabra Viranesque via Flickr / Creative Commons

Article Citation:

Hédoin, Cyril. 2021. “On Justifying Epistocratic Political Institutions: A Reply to Samuel Bagg.” Social Epistemology Review and Reply Collective 10 (6): 63-69. https://wp.me/p1Bfg0-5XS.

🔹 The PDF of the article gives specific page numbers.

This article replies to:

❧ Bagg, Samuel. 2021. “Reply to Cyril Hédoin’s ‘The “Epistemic Critique” of Epistocracy and Its Inadequacy’.” Social Epistemology Review and Reply Collective 10 (4): 31-36.

Articles in this dialogue:

❦ Reiss, Julian. 2019. “Expertise, Agreement, and the Nature of Social Scientific Facts or: Against Epistocracy.” Social Epistemology 33 (2): 183–92.

❦ Hédoin, Cyril. 2021. “The ‘Epistemic Critique’ of Epistocracy and Its Inadequacy.” Social Epistemology 1-13. doi: 10.1080/02691728.2021.1882609.

In my article, I argue that Reiss’s claims are plausible only under weak versions. As a result, they do not deliver the decisive blow to epistocracy that Reiss suggests. Moreover, once we acknowledge that social scientific knowledge is not as controversial as Reiss argues, another strand of the epistemic critique of epistocracy (the Epistemic Cost Argument) is weakened and ultimately inconclusive. Bagg essentially agrees with this last point but claims that once Reiss’s argument is more appropriately framed, my article fails to refute it. As a result, the epistemic critique of epistocracy cannot be disregarded after all.

Bagg’s reply makes a few specific and related claims that I shall sum up in the following ways:

(1) We should distinguish between how knowledge should be used for political governance and for political exclusion.

(2) The degree of confidence in the truth of specific scientific propositions required for the latter is higher than for the former.

(3) Choosing between candidates or parties in an election (task V) requires a broader set of knowledge KV than the knowledge KD for making decisions about policies in a specific domain D.

(4) There is no piece of knowledge in KV in which we have sufficient confidence that is sufficiently important or relevant to justify political exclusion in task V.

(5) At most, what my argument establishes is that someone lacking KD should be barred from decision making in domain D, but not from task V.

(6) Democracy should be our default position, and as long as the above claims have not been refuted, the epistemic critique of epistocracy holds.

I shall grant Bagg’s claims (1) and (2).

In a first part, I will dispute claims (3) and (4), thus casting doubts on claim (5). In a second part, I will explain why—irrespective of what one thinks of the previous claims—claim (6) is not appropriately framed.

Epistocracy and democracy are ideal types. In practice, political systems mix elements of both. I sketch a different rationale for considering expanding epistocratic features in political governance: to foster a political system that gives prominence more to the reasons underlying the citizens’ votes than their votes themselves.

Knowledge and Political Power

Epistocracy is loosely defined as the government by the wise. A more precise definition is however useful: a set of political institutions that distribute political power according to the degree of political and scientific knowledge individuals possess. When reflecting on political institutions, we may have at least two related but still separate kinds of considerations in mind: justificatory considerations and stability considerations. In this reply, I shall focus on the former, leaving the latter aside. Justificatory considerations are about the requirements that political institutions must meet for considering that the members of the polity have sufficient reasons to abide by them. Applied to the case of epistocracy, the justificatory problem turns into something like: do individuals have sufficient reasons to abide by political institutions that distribute political power based on political and scientific knowledge? I think that a necessary (and perhaps sufficient) condition for an affirmative answer is that what I shall call the “Epistemic Basis for Political Legitimacy” principle is accepted:

Epistemic Basis for Political Legitimacy (EBPL)—Political institutions are legitimate only if they distribute political power as a function of the distribution of knowledge in society and grant more power to agents who possess more knowledge.

Why the EBPL principle is a necessary condition for the justification of epistocratic institutions is obvious. Correlating the distribution of political power to the distribution of knowledge in the society is what distinguishes epistocracy from other political institutions, democratic or not.[1] Bagg’s claims (1)-(5) imply a rejection of the EBPL and thus of the possibility of justifying epistocracy.

Consider claims (1) and (2) first. Bagg contends that epistocratic institutions, whatever their specific forms, use knowledge as a criterion not only to govern but also to exclude. The difference may be explained in the following way. Governing based on knowledge means that policies and more generally collective choices should make the best use of the available knowledge. This may be viewed as a technocratic ideal that, as such, is compatible with democratic institutions.[2] Epistocracy goes further because it uses knowledge as a criterion to disenfranchise people. Voting rights are undoubtedly a component of political power. Does it mean however that those persons who are disenfranchised are excluded from the polity? The answer is positive if one accepts that:

Claim R: The right to vote is a necessary condition for political inclusion.

I think that claim R could be disputed in several ways. For instance, one may argue that as long as a person has the actual capacity to debate with others within fora or other political settings, then she is still a member of the polity. Moreover, it may also be noted that as far as political power is concerned, one’s vote has an insignificant weight, at least in nationwide elections. There may be more effective ways to acquire political power and thus to be part of the polity in a meaningful sense. I shall grant claim R however. Then, Bagg’s claims (1) and especially (2) can be both accepted. Given the implications of disenfranchisement for the political status of persons, it is reasonable to consider that accepting the EBPL principle requires setting the bar high regarding the confidence we must have in the truth of scientific propositions.

Denote K the set of scientific propositions that are considered as true by the relevant scientific communities with a sufficient degree of confidence. Denote KD the corresponding subset for any specific domain D. Finally, denote K the set of all these subsets.[3] The core of Bagg’s argument is that no single element in K is such that it warrants denying a person the right to vote (claim (4)). This is because choosing between candidates or parties (task V) is (or should be) based on a broader set of knowledge KV than the knowledge corresponding to any KD (claim (3)).

Put differently, the fact of voting relies (or should rely) on a more diverse set of considerations than knowledge about issues related to a specific domain. Voting for a candidate at the French or American presidential elections does not resume voting to the one whose program is consistent with any specific KD. This seems to be uncontroversial. Of course, an obvious answer is then to contend that KV = K. Is it plausible? I do not think so but, “fortunately” (from the epistocrat’s perspective), it does not necessarily imply that the EBPL should be rejected. Consider the kind of considerations that may account for the gap between KV and K. We may think first of a person’s knowledge of her interests. We may also think of what a person knows or believes about the personal attributes and competencies of the candidates, or if the candidates and their programs reflect values that one deems important. All of these considerations are empirically significant, as largely cited studies by Achen and Bartels (2016) and others indicate. However, what is at stake here is not an empirical but a normative issue: of all the elements that account for the gap between KV and K, which ones should be considered as relevant for the distribution of political power?

There is no doubt that people vote for a variety of reasons, including self-interested but also identity-based ones, and of course sometimes reasons related to prejudices devoid of any scientific foundation. Traditional defenses of epistocracy by Brennan (2016) and others contend that this gap has adverse consequences in terms of welfare. This “consequentialist” defense of epistocracy is however vulnerable to several objections.  For instance, epistemic democrats can argue that irrespective of this gap, there are still solid grounds to prefer democratic institutions because they are the best way to take advantage of how (scientific) knowledge is distributed in society. Beyond this debate, what is at issue here concerning Bagg’s claim (5) properly reformulated is whether it is legitimate to ground political exclusion on the fact that a person does not sufficiently master K, irrespective of the “knowledge” she has beyond K.

Two lines of reasoning may support a negative answer. The first is related to what I have called in my article the “Epistemic Costs Argument”. In particular, it is possible that by politically excluding part of the population because these persons do not sufficiently master K, epistocratic institutions contribute to the epistemic domination of some social groups over others. As I explain in part III of my article, this is a real possibility that should be taken seriously. The seriousness of this problem is however inversely correlated with the relative size of K. If very few scientific propositions qualify as moderately controversial in the relevant sense (i.e. the gap between K and KV is large), then the basis for exclusion is indeed small. If K is relatively large however, while still existent, the risk of epistemic domination should not be overestimated. Bagg challenges my various examples of propositions susceptible to belong to K (knowledge about the effects of various policies, understandings of the effects of various schemes of taxations…) on the ground that “[v]oting for candidates in an election, which is how the vast majority of people exercise political power, is a completely different task from making substantive decisions”. But the question then is, on which basis should a voter vote for a candidate/party?

Ultimately, to vote is to delegate decision-making on precisely those issues on which scientific knowledge exists. As I indicate above, the relevant comparison is not between the knowledge required for voting KV and every bit of scientific knowledge KD in a domain D. Voting in an election is to delegate political power to persons and institutions who will make decisions that encompass the whole set of domain-specific knowledge K. Then, the only plausible way to contend that the gap between KV and K is large is to argue that there are many other considerations based on—for instance—social identities, self-interest, and prejudices that should be normatively relevant when one is voting. This topic would need more than this reply to be discussed. I think however that it is sufficient to keep in mind that political decision-making is largely about how to produce (or not) public goods and public bads. Letting too much space to identity-based or self-interested considerations is in any case in contradiction with this requirement.

The second line of reasoning is more straightforward. One may argue, based on empirical evidence, that no matter how much a person masters K, her vote mostly depends on considerations that are unrelated to moderately controversial scientific knowledge anyways. As recently noted by Gibbons (forthcoming), psychological data on this question is ambiguous.[4] Let partially grant the point however, and admit that a person’s mastery of social scientific knowledge has a minor effect on her vote. I shall contend in the next section that there is still a way to justify epistocracy.

A Different Rationale for Epistocratic Institutions

In his reply, Bagg rightly notes that my argument relies on some sort of burden-shifting trick. Toward the end, he however argues that “there are many reasons not to take the justifiability of epistocracy as our default position… this puts the burden of proof on advocates of epistocracy to show that some deviation from this democratic baseline can in some cases be justified” (35). On what grounds should we settle for a baseline is of course a delicate issue. Opting for a historical baseline seems overly conservative. A baseline relying on ethical intuitions is probably more appropriate, provided that the latter are amenable to revision. It is arguable in any case that there is no obvious reason for taking “democracy” as the default position. As a matter of fact, modern democracies are not fully democratic in the sense that they also rely on several forms of political exclusion.

Exclusions based on age, nationality, and criminal record are the most obvious ones. Judicial reviews and the proliferation of “independent agencies” (including independent central banks) also are instances of partial political exclusion where the distribution of political power is explicitly non-democratic. The motivations for this partial exclusion are diverse, ranging from the preservation of fundamental liberal rights to the credibility and efficiency of public economic policies. The point is that what we call “democracy” refers to an ideal type that is to be distinguished from the concrete political institutions through which collective choices are made.

I emphasize this relatively obvious point to make it clear that claiming that democracy should be our default choice is ambiguous. Is it the ideal type that has yet to be implemented in practice, or some specific sets of political institutions which organize competitive elections but tolerate and even call for forms of political exclusion? If the latter, then the contrast with epistocracy is weaker and the choice of the default position less obvious. In particular, as witnessed by the epistemic turn in the literature on deliberative democracy, there is a growing awareness that epistemic considerations may be a basis for political legitimacy. In other words, depending on what we take as the appropriate baseline, the EBPL principle need not be regarded as running contrary to our default position.

Once this point is taken, are there any further reasons to increase the importance of epistocratic features in political governance based on the EBPL? The few defenses of epistocracy one can find in the recent literature are explicitly consequentialist. Epistocratic institutions would lead to better outcomes because they make collective choices depending on a more reliable informational basis. Call it the Outcome-Based Justification of Epistocracy—Epistocratic political institutions are justified because they would lead to better outcomes as evaluated by agreed-upon consequentialist criteria.

The assessment of this claim is made difficult because of its mostly counterfactual nature—no ideal epistocracy has been implemented on a large scale yet. Hence the question of who bears the burden of proof about showing that social sciences provide or not the kind of knowledge that would plausibly lead to better outcomes. A different and complementary approach to the justification of epistocracy can however proceed differently, without relying on any claim about outcomes. Let me call it the

Reason-Based Justification of Epistocracy—Epistocratic political institutions are justified because they would favor collective choices based on relevant reasons.

Both kinds of justification are consistent with the EBPL. Epistocratic institutions allocate political power as a function of the distribution of knowledge in society. The outcome-based approach justifies this allocation because it would presumably lead to better outcomes. The reason-based approach justifies it because it would presumably favor collective decision-making that relies on “better”, “more relevant”, “less controversial” reasons. Consider Reiss’s (2019) first claim “There is no such thing as superior political judgment” in its weak version. That means that we accept the distinction between technical and political judgments but also admit that the latter can be partially ordered based on the truth value of the technical judgments on which they rely. Assume that the truth value of any particular technical judgment JT is settled by the relevant piece of scientific knowledge KD in some domain D. Now, consider the following thesis:

Partial Ordering of Political Judgments (POPJ)—Among two political judgments JP and JP’ over a specific issue I, JP is better than Jpif, everything else equals, JP but not JP is consistent with a technical judgment JT which is regarded as true according to the relevant piece of scientific knowledge KD.

POPJ claims that we can discriminate between political judgments whether they are consistent with true technical judgments. For the moment, suppose that we accept POPJ. That means that a political judgment on a political issue (e.g. on whether we should open borders) is better than another one on the same issue if it is consistent with the relevant scientific knowledge (e.g. opening borders does not systematically produce downward pressures on wages). This also applies to larger political issues such as for whom to vote in an election. The political judgment “X has the best program” is better than “Y has the best program” if, everything else equals, the former is more consistent with scientific knowledge than the latter. Presumably, epistocratic political institutions would favor political judgments that are consistent with true technical judgments. On the other hand, the ordering of political judgments would be radically incomplete, especially for large scale political issues, like voting for a candidate. This is because political judgments are underdetermined by technical judgments, or to use Bagg’s notation, the knowledge KV for accomplishing task V of voting cannot be reduced to domain-specific scientific knowledge KD, or even to global scientific knowledge K.

The question is of course whether POPJ should be accepted. Bagg, Reiss, and other critics of epistocracy may respond negatively. They could contend that for most political issues, there is just no piece of scientific knowledge that qualifies as relevant because it is too controversial. But then we are back at Reiss’s second claim that “there is no such thing as uncontroversial social scientific knowledge” in its strong version. And here, I stand by my claim (which Bagg does not dispute in his reply) that in this case, consistency requires to go further and to question the place of (social) science in secular societies. Of course, that does not settle the issue in favor of epistocracy. POPJ is also consistent with deliberative democratic institutions that, while not relying on political exclusion, implement collective decision processes that favor the emergence of a consensus based on shared and presumably relevant reasons.

The recent French experience of the “Convention Citoyenne pour le Climat” provides a good illustration of an institution promoting social decision-making based on comprehensive scientific evidence and informed deliberation, without relying on extra-democratic forms of political exclusion.[5] But the margin between this kind of democratic institutions and epistocratic ones that would implement enfranchisement lotteries with competency tests is actually thin. Here then lies the best justification of epistocracy: to encourage public deliberation and social decision-making that rely on the best scientific evidence and, more generally, on reasons that are most likely to be shared in a population.

Author Information:

Cyril Hédoin, cyril.hedoin@univ-reims.fr, is professor of economics at the University of Reims Champagne-Ardenne (France). His academic work is at the crossroads of philosophy and economics and focuses more recently on topics articulating normative economics and political and moral philosophy. He has recently published articles in economics or philosophy journals like Economics and Philosophy, Erkenntnis and Philosophy of the Social Sciences.


Achen, Christopher H., and Larry M. Bartels. 2016. Democracy for Realists: Why Elections Do Not Produce Responsive Government. Princeton: Princeton University Press.

Bagg, Samuel. 2021. “Reply to Cyril Hédoin’s ‘The “Epistemic Critique” of Epistocracy and Its Inadequacy’.” Social Epistemology Review and Reply Collective 10 (4): 31-36.

Brennan, Jason. 2016. Against Democracy. Princeton: Princeton University Press.

Friedman, Jeffrey. 2020. Power without Knowledge: A Critique of Technocracy. New York: OUP USA.

Gibbons, Adam F. forthcoming. “Is Epistocracy Irrational?” Journal of Ethics and Social Philosophy.

Hédoin, Cyril. 2021. “The ‘Epistemic Critique’ of Epistocracy and Its Inadequacy.” Social Epistemology 1-13. doi: 10.1080/02691728.2021.1882609.

Reiss, Julian. 2019. “Expertise, Agreement, and the Nature of Social Scientific Facts or: Against Epistocracy.” Social Epistemology 33 (2): 183–92.

[1] Of course, note that that the EBPL principle does not say that political legitimacy is only based on knowledge. Other legitimacy conditions may be imposed, e.g. that individuals have voluntarily consented to the institutions.

[2] See Friedman (2020) for a study of the relationships between democracy and technocracy.

[3] I leave open here how much coarse/fine K should be. Note however that it is unlikely that K is a partition of K, i.e. domains of knowledge plausibly intersect.

[4] All hinge on the way “knowledgeable persons” are defined and identified. The empirical literature is ambiguous and tends to focus on measures of “political knowledge”. Then, it seems that persons with a higher degree of political knowledge are also more prone to motivated reasoning and confirmation bias. However, it is not obvious that these are these persons who would be granted more political power under epistocratic institutions (Gibbons, forthcoming).

[5] Except for the initial random selection of the participants. For more information, see https://www.conventioncitoyennepourleclimat.fr/.

Categories: Critical Replies

Tags: , , , , , , , , ,

Leave a Reply